An AI chatbot can generate personality tests and predict people's responses before they take them, according to new research. Scientists developed a method for generating personality assessment questionnaires with ChatGPT from any source text, applying it to the fifth edition of the American Psychiatric Association's manual (DSM-5) and, as a deliberately unconventional example, an astrology textbook. The findings, published in the journal iScience, showed that ChatGPT could not only create and validate the questionnaires but also accurately predict responses before the surveys were administered. ChatGPT and other large language models (LLMs) are trained on the internet by compiling trillions of human language data inputs from websites and social media. Study lead author Rotem Monsa, a doctoral student at the Hebrew University of Jerusalem in Israel, said personality traits are reflected in language, so LLMs may have learned the structure of human personality as a byproduct of their training. The researchers used GPT-4 to generate two questionnaires, first using excerpts from the DSM-5 as source text and, as a control, an astrology textbook. For the DSM-5, ChatGPT generated statements based on descriptions of personality disorders, such as "often suspects others' motives" or "finds it easy to trust people." The astrology questionnaire generated similar statements based on zodiac signs. Monsa noted the texts sit on opposite ends of a spectrum in terms of scientific grounding, with the DSM-5 refined through decades of clinical research and the astrology text culturally based but not scientifically valid.